We start by considering a conceptual framework in which a farmer decides whether to expand farmland beyond his landholding, given a set of parameters regarding expected prices, policies and technology. This framework suggests that conservation policies should be binding whenever the farmer faces a tight land constraint. Its two main implications guide our empirical analysis of the impact of conservation policies: (i) agricultural prices must be controlled for when evaluating policy; and (ii) the farmer’s response to policy stringency depends on the tightness of land constraints. The second implication introduces cross-sectional variation in response to policy among different municipalities.
We apply this framework to evaluate the effectiveness of the conservation policies introduced starting in 2004 and 2008 considering a sample of municipalities from selected Legal Amazon states in the 2002 through 2009 period. We use the normalized deforestation increment for each municipality and year as the dependent variable. Exogenous cross-sectional variation in price indices is obtained with the use of agricultural prices from a southern Brazilian state weighted by the relative importance of different agricultural products at the local level before the period of analysis. The set of conservation policies are represented by two dummy variables - one for each policy turning point - and their interaction with a measure of the tightness of land constraints in each municipality. The baseline proxy for the tightness of land constraint is a measure of the share of land unavailable for agriculture relative to the total area of each municipality.
We first show that deforestation rates have been responsive to agricultural output prices, even after controlling for year and municipality fixed effects. Results suggest that crop and cattle prices have different dynamic relationships with deforestation. We then perform policy evaluation controlling for as much of the price effects as possible, including municipality-specific time trends. The empirical results indicate that the conservation policies associated with the two turning points were effective in curbing deforestation rates in Brazil. In counterfactual simulations we estimate that these policies avoided 62,100 km2 of deforestation, or 52.1% of the total deforestation that would have occurred in the sample in the 2005 through 2009 period had the policies adopted beginning in 2004 and 2008 not been introduced. Using the conversion factors of 10,000 tons of C per square kilometer and 5 US dollars per ton of CO2, this avoided deforestation is equivalent to an avoided loss of 621 million tons of stored C, or 2.3 billion tons of stored CO2, which is valued at 11.5 billion US dollars.